Video Block Split Prediction for Complex Scene Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video data compression methods struggle with predicting images when object and background images overlap or have significant illuminance changes, leading to inefficient block splitting and compression performance.
Innovation Solution
A video encoding/decoding method that splits blocks into various shapes, including curves and lines, to improve prediction accuracy and compression efficiency by generating predicted blocks and residual blocks, and encoding/decoding these using transformers and quantizers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a block is split into square shapes according to conventional H.264/AVC, then the block structure is simple and easy to process, but prediction efficiency deteriorates when object and background images overlap or illuminance changes significantly
Solution Approach 1:
The patent divides a block into multiple sub-blocks (e.g., 4×4, 8×4, 4×8, or 8×8 configurations) and applies different prediction modes to different sub-blocks. This segmentation allows the system to handle complex scenes with overlapping objects or illuminance changes by treating different regions independently, thereby improving prediction efficiency while maintaining manageable complexity through systematic subdivision.
Solution Approach 2:
The patent introduces dynamic block splitting where the number and arrangement of sub-blocks can vary based on the content being encoded. Instead of fixed square partitions, the system dynamically adjusts the partitioning structure (e.g., changing from 4 equal sub-blocks to asymmetric configurations) to adapt to different scene complexities, thus improving prediction accuracy for diverse visual content.
2Reliability
If a block is split into asymmetric rectangular shapes, then compression is improved when object and background have asymmetric layout, but block splitting becomes inaccurate for arbitrary shapes other than rectangular
Solution Approach 1:
The patent segments blocks into configurable sub-blocks that can be arranged in various patterns (4×4, 8×4, 4×8, 8×8) to approximate arbitrary shapes. By dividing complex shapes into multiple simpler sub-blocks, the system achieves accurate representation of non-rectangular objects and backgrounds while maintaining the simplicity of rectangular sub-block processing.
Solution Approach 2:
The patent extends the partitioning approach by allowing different numbers of rows and columns in sub-block arrangements, effectively adding dimensional flexibility. Instead of being limited to simple horizontal or vertical splits, the system can create grid-based partitions that better conform to arbitrary object shapes, improving adaptability without sacrificing rectangular sub-block simplicity.
3Measurement precision
If arbitrary direction splitting is implemented using length and angle information, then splitting accuracy for arbitrary shapes improves, but data transmission amount increases and compression efficiency decreases due to quantization losses
Solution Approach 1:
The patent segments the block into a finite set of predefined sub-block configurations (4×4, 8×4, 4×8, 8×8 arrangements) rather than allowing arbitrary directional splits. This segmentation approach achieves sufficient splitting accuracy for most video content by using a manageable number of discrete partition types, avoiding the need to transmit continuous angle and length parameters.
Solution Approach 2:
The patent uses a limited set of standardized sub-block partition patterns that can be efficiently encoded and decoded. Instead of transmitting detailed arbitrary split information (which would require significant data), the system selects from a small repertoire of pre-defined partition types, significantly reducing the bit rate required for partition information while maintaining adequate splitting accuracy for video compression purposes.
Data Source
AI summary
A video encoding/decoding method and apparatus using split prediction is provided.The video encoding/decoding method and apparatus using split prediction splits a block into various partition shapes in addition to a conventional standardized right-angled quadrangular shape so as to effectively predict an image of a current block from a previous image and thus, may more improve the performance of video data compression and may achieve superior reconstructed image quality.